TDC - Trilha Machine Learning - O que sabemos de voce por meio de PLN e ML?
?
2 likes?249 views
This document is an agenda for a presentation on forensic analysis of messages on social networks. The agenda includes sections on theoretical background, general methodology, performance evaluation, and introduction. It aims to analyze what can be known about a user through natural language processing and machine learning of their social media messages.
![PR?-PROCESSAMENTO
[('Como', u'KS'), ('dizem', u'V'), ('meus', u'PROADJ'), ('sobrinhos',
u'N'), ('de', u'PREP'), ('Twitter', None), (',', u','), ('rimos', None),
('litros', u'N'), ('.', u'.'), ('Embora', u'KS'), ('eu', u'PROPESS'), ('nao',
u'ADV'), ('entenda', u'V'), ('como', u'PREP'), ('se', u'PROPESS'),
('possa', u'VAUX'), ('medir', u'V'), ('o', u'ART'), ('volume', u'N'),
('espacial', u'ADJ'), ('de', u'PREP'), ('risos', u'N'), ('.', u'.'), ('Mas',
u'KC'), ('tudo', u'PROSUB'), ('bem', u'ADV')]
TWEET PR?-PROCESSAMENTO](https://image.slidesharecdn.com/tdc-machinelearning-joseahirton-180720113241/85/TDC-Trilha-Machine-Learning-O-que-sabemos-de-voce-por-meio-de-PLN-e-ML-14-320.jpg)
TDC - Trilha Machine Learning - O que sabemos de voce por meio de PLN e ML?
- 2. O R I E N T A D O R P r o f . D r . L e a n d r o N u n e s d e C a s t r o P r o f . D r . R o d r i g o P a s t i C O - O R I E N T A D O R L a b o r a t ó r i o d e C o m p u t a ? ? o N a t u r a l - L C o N P r o g r a m a d e P ó s - G r a d u a ? ? o e m e n g e n h a r i a E l é t r i c a e C o m p u t a ? ? o U n i v e r s i d a d e P r e s b i t e r i a n a M a c k e n z i e Análise forense de mensagens em rede social - O que podemos saber de você por meio de PLN e ML?
- 3. AGENDA 2 UCTION REFERENCIAL TE?RICO METODOLOGIA GENEC AVALIA??O DE DESEMPENHO INTRODU??O
- 4. INTRODU??O
- 10. Your love got me You hit me so hard with your love. He knocked me out, he stunned me It came out, I went to the canvas, it was K.O. I've always been a warrior, but first- I saw myself helpless, heart lost the fight, yes Goodbye drunk, single life, I want Friday To be with you in my bed, together, coladin Kiss me all night, sexy in the bathtub I'm going to make you tired, I want to start from the beginning And look, I'll look at you I'll tell you, I went to the canvas with your Your love got me You hit me so hard with your love. He knocked me out, he stunned me It came out, I went to the canvas, it was K.O. I've always been a warrior, but first- I saw myself helpless, heart lost the fight, yes Goodbye drunk, single life, I want Friday I saw myself helpless, heart lost the fight, yes Goodbye drunk, single life, I want Friday I saw myself helpless, heart lost the fight, yes Goodbye drunk, single life, I want Friday Your love got me You hit me so hard with you He knocked me out, he stun It came out, I went to the canvas I've always been a warrior, b I saw myself helpless, heart lost t Goodbye drunk, single life, I w To be with you in my bed, togeth Kiss me all night, sexy in the I'm going to make you tired, I w from the beginning And look, I'll look at yo I'll tell you, I went to the canvas Your love got me You hit me so hard with you He knocked me out, he stun It came out, I went to the canvas I've always been a warrior, b I saw myself helpless, heart lost t Goodbye drunk, single life, I w I saw myself helpless, heart lost t genero comunica??o cultura EXPRESS?O DE G?NERO
- 11. ARQUITETURA COLETAR BASE DE DADOS PR?-PROCESSAMENTO EXTRA??O DE META-ATRIBUTOS SELE??O DE META-ATRIBUTOSCLASSIFICA??ORESULTADO
- 12. PR?-PROCESSAMENTO BASE DE DADOS PR?-PROCESSAMENTO TOKENIZA??O POS TAGGING
- 14. PR?-PROCESSAMENTO [('Como', u'KS'), ('dizem', u'V'), ('meus', u'PROADJ'), ('sobrinhos', u'N'), ('de', u'PREP'), ('Twitter', None), (',', u','), ('rimos', None), ('litros', u'N'), ('.', u'.'), ('Embora', u'KS'), ('eu', u'PROPESS'), ('nao', u'ADV'), ('entenda', u'V'), ('como', u'PREP'), ('se', u'PROPESS'), ('possa', u'VAUX'), ('medir', u'V'), ('o', u'ART'), ('volume', u'N'), ('espacial', u'ADJ'), ('de', u'PREP'), ('risos', u'N'), ('.', u'.'), ('Mas', u'KC'), ('tudo', u'PROSUB'), ('bem', u'ADV')] TWEET PR?-PROCESSAMENTO
- 15. EXTRA??O DE META-ATRIBUTOS CARACATERES E SINTAXE CARACTERES E SINTAXE PALAVRAS ESTRUTURA TEXTUAL MORFOLOGIA PSICO-LINGU?STICA # TOTAL CARACTERES : ; ...K DE YULE R DEHONORE S DE SICHEL D DE SIMPSON ENTROPIA ORGANIZA??O DE CONTE?DO MARCAS EMOCIONAIS, PSICOL?GICAS E COGNITIVAS 16 28 10 06 68 128
- 16. EXTRA??O DE META-ATRIBUTOS 134 0.7761 0.0298 0.0 0.1716 tweet vetor de característica white/Cespa?o em branco/C dígitos/C A-Z/C a-z/C C VALORMETA-ATRIBUTOS
- 17. CLASSIFICA??O
- 20. AVALIA??O DE DESEMPENHO 100 tweets M?TODO META-ATRIBUTOS MAIS SIGNIFICANTEST?CNICA TWEETSCONCATENADOS(100) GANHODE INFORMA??O
- 21. AVALIA??O DE DESEMPENHO 500 tweets M?TODO T?CNICA META-ATRIBUTOS MAIS SIGNIFICANTES TWEETSCONCATENADOS(500) GANHODE INFORMA??O
- 22. CONCLUS?O ~70% ?RVORE DE DESI??O ~75% M?QUINA DE VETOR SUPORTE (SVM)
- 23. AVALIA??O DE DESEMPENHO RESULTADOS CONJUNTO DE META-ATRIBUTOS SINTAXE E CARACTERES PALAVRAS ESTRUTURA TEXTUAL MORFOLOGIA TEXTUAL PSICO-LINGU?STICA TUDO
- 24. CONCLUS?O Análise forense de mensagens an?nimas Ferramenta de recomenda??o para segmenta??o E-mail makerting Constru??o de perfil de usuário
- 25. “Trabalho dedicado a Guilherme Henrique Ferreira Nunes Lopes, meu irm?o (in memoriam), um aventureiro que com sua simplicidade me ensinou a respeitar a natureza das coisas. Que saibamos respeitar cada um, assim como o surfista aprendeu a respeitar o mar, a onda, o infinito em cada um.”